Hunting for the top partner in the littlest Higgs model with T parity at the CERN LHC

PHYSICAL REVIEW D 75, 556 (27) Hunting for the top partner in the littlest Higgs model with T parity at the CERN LHC Shigeki Matsumoto,* Mihoko M. Nojiri, and Daisuke Nomura Theory Group, KEK, 1-1 Oho, Tsukuba, 35-81, Japan (Received 5 January 27; published 12 March 27) We study the littlest Higgs model with T parity at the CERN LHC through pair productions of the T-odd top quark partner (T ) which decays into the top quark and the lightest T-odd particle. We identify the region of parameters favored by the electroweak and cosmological considerations. The signal and background events are simulated with fast detector simulation to study the discovery potential at the LHC. We find that the hemisphere analysis recently proposed by the CMS Collaboration is very useful to separate the signal from the t t background. We discuss the observability of the top tagged signal in the effective mass (M eff ) versus the transverse missing energy (E Tmiss ) plane. We show that, for all our sample parameter sets with M T 9 GeV, the excess of the signal over the background can be visible as a bump structure in the E Tmiss distribution for 5 fb 1 at relatively high M eff intervals. DOI: 1.113/PhysRevD.75.556 PACS numbers: 14.8.Cp, 12.6.Cn, 14.65.Ha *Email address: smatsu@post.kek.jp Email address: nojiri@post.kek.jp Email address: dnomura@post.kek.jp I. INTRODUCTION The Higgs sector in the standard model (SM) receives large quadratic mass corrections from top and gauge boson loop diagrams. New symmetries involving top-higgs and gauge-higgs sectors below O (1 TeV) are proposed to remove the corrections. One of the physics targets of the ATLAS [1] and CMS [2] experiments at the CERN LHC is to find new particles predicted by such symmetries. The most important new physics candidate is the minimal supersymmetric standard model (MSSM)[3]. This model predicts quark and lepton partners with spin (squark and slepton) and those of gauge and Higgs bosons with spin 1=2 (gauginos and Higgsinos). Thanks to the cancellation among boson and fermion loop diagrams, the quadratic corrections to the Higgs mass term completely vanish. At the LHC, strongly interacting supersymmetric particles such as a squark and a gluino will be copiously produced. They decay into relatively light electroweak (EW) superpartners, such as the chargino, neutralino, and slepton. The lightest supersymmetric particle (LSP) is stable due to the R parity of the model. The decay products of the squark and gluino contain at least one LSP. The LSP escapes from the detector without any energy deposit, giving a missing momentum signature to the events. The signature of supersymmetric particles has been studied intensively by many groups [1,2,4]. Alternative scenarios which do not rely on supersymmetry to cancel the quadratic divergent corrections have been proposed and studied. The little Higgs model [5] is one of these alternatives. In this model, the Higgs boson is regarded as a pseudo Nambu-Goldstone boson, which originates from the spontaneous breaking of a global symmetry at a certain high scale, and the global symmetry protects the Higgs mass from the quadratic radiative corrections. The simplest version of the model is called the littlest Higgs model [6]. The global symmetry of the model is SU(5), which is spontaneously broken into SO(5). Part of the SU(5) symmetry is gauged, and the gauge symmetry is SU 2 1 SU 2 2 U 1 1 U 1 2. The top sector is also extended to respect the part of the global symmetry. This model, however, predicts a large correction to the EW observables because of the direct mixing between heavy and light gauge bosons after the EW symmetry breaking. The precision EW measurements force the masses of heavy gauge bosons and top partners to be O (1 TeV), reintroducing the fine-tuning problem to the Higgs mass [7]. A solution of the problem is the introduction of T parity to the model which forbids the mixing [8]. This is the symmetry under the transformations SU 2 1 $ SU 2 2 and U 1 1 $ U 1 2. All heavy gauge bosons are assigned a T-odd charge, while the SM particles have a T-even charge. The matter sector should be extended so that T-odd partners are predicted. The lightest T-odd particle is the heavy photon, which is stable and becomes a candidate for dark matter [9 11]. Starting from a different underlying theory, the littlest Higgs model with T parity ends up predicting a similar phenomenology to that of the MSSM. We view that this is indispensable for the model to remove the hierarchy problem. One needs a new symmetry to protect the Higgs mass to reduce the quadratic divergence of the theory in a meaningful manner. The symmetry must involve both the top quark and the gauge sectors, because the Higgs couplings to these particles are the dominant source of the divergences. Unless some parity is not assigned to the gauge partner, large corrections to the EW observables are expected, and those are not acceptable after the CERN LEP era. Some of the top and gauge partners are required in the parity odd sector of the model, while SM particles are in the parity even sector. Notably, if this parity is exact, the lightest parity odd particle is stable; therefore the picture is consistent with the existence of the dark matter in our Universe. Finally, the production of the top partners and 155-7998=27=75(5)=556(14) 556-1 27 The American Physical Society

MATSUMOTO, NOJIRI, AND NOMURA PHYSICAL REVIEW D 75, 556 (27) their decays into the top quark and an invisible particle are predicted as the signal of beyond the standard model at the LHC. For the case of the MSSM, the pair production of the scalar top is not detectable, because the production cross section is too small. Supersymmetry is instead detected by the production of gluinos and squarks in the first generation, and the scalar top may appear as the decay product of the gluino [12]. On the other hand, the cross section of the top partner pair production in the littlest Higgs model with T parity is about 1 times larger than that of the scalar top, so that they may be detectable at the LHC [13]. The purpose of this paper is to provide a realistic estimate for the detection of the fermionic top partner at the LHC. This paper is organized as follows. In Sec. II, we review the littlest Higgs model with T parity, focusing on the top sector. In Sec. III, we summarize the electroweak and dark matter constraints on the model. We find that the lower limit of the top partner mass is about 6 GeV, and the lightest T-odd particle is always much lighter than the top partner. The signal at the LHC is therefore two top quarks with significant transverse momentum and missing energy. In Secs. IVand V, we discuss the top partner signature at the LHC. Reconstructing a top quark in the event is essential to identify the top partner. We describe the method to reconstruct the top quark in Sec. IV. We apply a hemisphere analysis to the signal reconstruction, and study the reconstruction efficiency of the top quark for both signal and t t background. In Sec. V, we discuss the basic cuts to reduce the top quark background. The highest S=N ratio is obtained in the region where the effective mass is around twice the top partner mass. We also show the numerical results of our simulation study and find that the top partner signature would be significant over the background if the mass of the top partner is less than 1 TeV. Section VI is devoted to discussion and comments. II. THE LITTLEST HIGGS MODEL WITH T PARITY In this section, we briefly review the littlest Higgs model with T parity, focusing on the top sector of the model. The constraints on the model from WMAP observations and electroweak precision measurements will be discussed in the next section. For general reviews of the little Higgs models and their phenomenological aspects, see Refs. [14,15]. A. Gauge-Higgs sector The littlest Higgs model [6] is based on a nonlinear sigma model describing an SU 5 =SO 5 symmetry breaking. The nonlinear sigma field,, is given as e 2i =f ; (1) where f is the vacuum expectation value associated with the breaking and is expected to be O 1 TeV. The Nambu- Goldstone (NG) boson matrix and the direction of the breaking are p H= 1 2 B p H y = p @ 2 H T = p 2 y H = C A; 2 1 (2) 1 @ 1 A; 1 where we omit would-be NG fields in the matrix. An SU 2 U 1 2 subgroup in the SU(5) global symmetry is gauged, which is broken down to the diagonal subgroup identified with the SM gauge group, SU 2 L U 1 Y. Because of the presence of the gauge interactions (and Yukawa interactions introduced in the next subsection), the SU(5) global symmetry is not exact, and particles in the matrix become pseudo NG bosons. Fourteen ( 24 1) NG bosons are decomposed into representations under the electroweak gauge group as 1 3 2 1=2 3 1. The first two representations are real, and become longitudinal components of heavy gauge bosons when the SU 2 U 1 2 is broken down to the SM gauge group. The other scalars in the representations 2 1=2 and 3 1 are the complex doublet identified with the SM Higgs field [H in Eq. (2)] and a complex triplet Higgs field [ in Eq. (2)], respectively. The kinetic term of the field is given as where L f2 8 Tr D D y ; (3) D @ i X2 j 1 g j W j W T j g j B j B T j : (4) Here, W j Wj aqa j (B j B j Y j ) is the SU 2 j (U 1 j ) gauge field and g j g j is the corresponding gauge coupling constant. The generators Q j and Y j are Q a 1 1 a 1 @ A; 2 Y 1 diag 3; 3; 2; 2; 2 =1; 1 Q a 2 1 @ A; 2 a Y 2 diag 2; 2; 2; 3; 3 =1; where a is the Pauli matrix. In terms of the above fields, the symmetry under T parity [8] is defined as the invariance of the Lagrangian under the transformation: (5) 556-2

HUNTING FOR THE TOP PARTNER IN THE LITTLEST... PHYSICAL REVIEW D 75, 556 (27) W1 a $ Wa 2 ; B 1 $ B 2 ; $ or $ ~ (6) y ; where diag 1; 1; 1; 1; 1. As a result of the symmetry, the gauge coupling p g 1 (g 1 ) must be equal to g 2 (g 2 ), namely, g 1 g 2 2 g (g 1 g 2 p 2 g ), where g (g ) is nothing but the coupling constant of the SM SU 2 L (U 1 Y ) gauge symmetry. The Higgs potential is generated radiatively [6,9], V H; f 2 Tr y 2 H y H 4 Hy H 2 : (7) The main contributions to 2 come from the logarithmic divergent corrections at 1-loop level and quadratic divergent corrections at 2-loop level. As a result, 2 is expected to be smaller than f 2. The triplet Higgs mass term, on the other hand, receives quadratic divergent corrections at 1- loop level, and therefore is proportional to f 2. The quartic coupling is determined by the 1-loop effective potential from the gauge and top sectors. Since both and depend on the parameters at the cutoff scale, we treat them as free parameters in this paper. Next, we discuss the mass spectrum of the gauge and Higgs bosons. This model contains four kinds of gauge fields, W1 a, Wa 2, B 1, and B 2, in the electroweak gauge sector. The linear combinations W a W1 a p p Wa 2 = 2 and B B 1 B 2 = 2 correspond to the SM gauge bosons for the SU 2 L and U 1 Y symmetries. The other linear combinations, WH a Wa 1 p p Wa 2 = 2 and BH B 1 B 2 = 2, are additional gauge bosons, which acquire masses of O f through the SU 5 =SO 5 symmetry breaking. After the electroweak symmetry breaking, the neutral components of WH a and B H are mixed with each other, and form mass eigenstates A H and Z H. The masses of the heavy bosons are m AH :45g f; m ZH m WH gf: (8) The mixing angle H between W 3 H and B H is considerably suppressed. It is given by tan H g v 2 = 4gf 2, where v ( 246 GeV) is the vacuum expectation value of the Higgs field. Thus the dominant component of A H is B H. Finally, the mass of the triplet Higgs boson is given by m 2 f 2 2m 2 h f2 =v 2, where m h is the SM Higgs boson mass. Under T parity, the new heavy gauge bosons and the triplet Higgs boson behave as T-odd particles, while SM particles are T-even. As shown in Eq. (8), the heavy photon is considerably lighter than the other T-odd particles. Stability of A H is guaranteed by T-parity conservation, and it becomes a candidate for dark matter. B. Top sector To implement T parity, two SU(2) doublets, q 1 and q 2, are introduced for each SM fermion doublet. Furthermore, two vectorlike singlet top partners, U 1 and U 2, are also introduced in the top sector in order to cancel large radiative corrections to the Higgs mass term. Since we are interested in top partner production at the LHC, only the top sector is discussed here. For the other matter sectors, see Refs. [9,16]. The quantum numbers of the particles in the top sector under the SU 2 U 1 2 gauge symmetry are shown in Table I. All fields in the table are triplets under the SM SU 3 c (color) symmetry. Using these fields, the Yukawa interaction terms which are invariant under T parity and gauge symmetries turn out to be L t p 1f 2 ijk xy Q 1 i jx ky Q 2 i jx ~ ky ~ u R 2 2 f U L1 U R1 U L2 U R2 H:c:; (9) where Q i @ q i U Li 1 A ; qi 2 u Li b Li : (1) The indices i, j, k run from 1 to 3 whereas x, y 4,5.The coupling constant 1 is introduced as the top Yukawa coupling, while 2 f gives the vectorlike mass term for the singlet fields. Under T parity, q i and U i transform as q 1 $ q 2 and U 1 $ U 2. Therefore, the T-parity eigenstates are given by q p 1 q 1 q 2 ; 2 U R p 1 U R1 U R2 : 2 U L p 1 U L1 U L2 ; 2 (11) In terms of these eigenstates, the mass terms for these quarks are written as follows, L mass 1 f U L v u L u R 2 f U L U R U L U R H:c: (12) The remaining T-odd fermion, q, acquires mass by introducing an additional SO(5) multiplet, transforming nonlinearly under the SU(5) symmetry. Therefore, the mass term of the q quark does not depend on 1 and 2. In this paper, we assume that the q quark is heavy compared to other top partners, and do not consider its production at the LHC. For the q quark phenomenology, see Refs. [16,17]. The T-even states u and U form the following mass eigenstates: TABLE I. sector. The SU 2 U 1 2 charges for particles in the top q 1 2; 1=3; 1; 2=15 q 2 1; 2=15; 2; 1=3 U L1 1; 8=15; 1; 2=15 U L2 1; 2=15; 1; 8=15 U R1 1; 8=15; 1; 2=15 U R2 1; 2=15; 1; 8=15 u R 1; 1=3; 1; 1=3 556-3

MATSUMOTO, NOJIRI, AND NOMURA PHYSICAL REVIEW D 75, 556 (27) t L cos u L sin U L ; T L sin u L cos U L ; t R cos u R sin U R ; T R sin u R cos U R ; (13) where sin 1 = 2 1 2 2 1=2 and sin 2 1 v= 2 1 2 2 f. The t quark is identified with the SM top quark, and T is its T-even heavy partner. On the other hand, the T-odd fermions U L and U R form a Dirac fermion, T. The masses of these quarks are given by m t 1 2 v q q ; m T 2 2 1 1 2 2f; 2 2 (14) m T 2 f: It is worth noting that the T-odd states do not participate in the cancellation of quadratic divergent corrections to the Higgs mass term. The cancellation is achieved only by loop diagrams involving t and T quarks. III. WMAP AND EW PRECISION CONSTRAINTS In the previous section, four parameters were introduced in the gauge-higgs and top sectors in addition to the gauge coupling constants (g and g ) and the vacuum expectation value of the Higgs field (v). These are m h, f, 1, and 2. Since the top quark mass is determined by the combination of v, 1, and 2, the number of undetermined parameters is three. These parameters can be expressed by m h, m AH, and m T. In this section, we see that m h and m AH are directly related to each other thanks to the precise data of the WMAP observations. Therefore, it is possible to write down the Higgs mass as a function of the dark matter mass, m h m h m AH. The electroweak precision measurements provide a further constraint on m h and m T. They gives a lower bound on m T and a large mass difference between T and A H. A. Constraint from WMAP observation First, we consider the WMAP constraint on the littlest Higgs model with T parity. The dark matter, A H, annihilates mainly into weak gauge bosons, W W and ZZ, through diagrams in which the Higgs boson propagates in the s channel. Once we calculate the annihilation cross section, the thermal relic abundance of the dark matter is obtained by solving the Boltzmann equation. For the detailed calculation of the abundance in this model, see Refs. [9 11].To good accuracy,the dark matter abundance can be written as 1pb c DM h 2 8:4 1 2 ; (15) v rel where c is the speed of light, and v rel is the relative velocity between initial dark matters. Since the product of the cross section and the relative velocity, v rel, and hence the dark matter abundance, DM h 2, depend only on m AH and m h, the WMAP observation, DM h 2 :111, gives a relation between these parameters. In Fig. 1, the thermal relic abundance of dark matter is depicted as a contour plot in the m AH ;m h plane. The thin shaded area is the allowed region from the WMAP observation at 2 level, :94 < DM h 2 < :129 [18]. In the figure, there are two branches: the upper branch (U branch) and the lower branch (L branch). These branches are sometimes called low and high regions, respectively [11]. In the U branch, the Higgs boson mass is larger than twice the dark matter mass, m h > 2m AH, while m h < 2m AH in the L branch. The Higgs mass is precisely determined by the dark matter mass up to a twofold ambiguity by imposing the WMAP constraint. m (GeV) h B. Constraint from electroweak precision measurements Next, we discuss the constraint from electroweak precision measurements. New physics contributions to electroweak observables come from radiative corrections, since there is no tree-level effect due to T parity. The constraint is sensitive to the masses of the Higgs boson and the top partner. Since the Higgs mass is directly related to the dark matter mass through the WMAP constraint, the electroweak precision constraint on m h and m T is translated into one on m AH and m T. In order to obtain the constraint, we follow the procedure in Ref. [19] using the S, T, and U parameters [2]. In that paper, it has been shown that main contributions to the parameters come from the top sector and the custodialf (TeV).6.8 1 2 1 5 1 1 U branch.1 L branch 1 2 3 m (GeV) A H.1 1.5.5 FIG. 1. Contour plot of the thermal relic abundance of the dark matter, DM h 2, in the m AH ;m h plane. The thin shaded area is the allowed region from the WMAP observation at 2 level, :94 < DM h 2 < :129. 1 556-4

HUNTING FOR THE TOP PARTNER IN THE LITTLEST... PHYSICAL REVIEW D 75, 556 (27) U branch L branch 3 3 (GeV) 25 Excluded 25 2 2 68% 99% (GeV) H H A A Excluded 68% m m 15 15 99% 68% 1 1 4 6 8 1 12 14 4 6 8 1 12 14 99% m T (GeV) m T (GeV) 68% FIG. 2. Constraints on m T and m AH at 68% and 99% confidence levels. At each point in these figures, the Higgs mass is determined to satisfy the WMAP constraint on the U branch (left panel) and L branch (right panel). symmetry violating effect from heavy gauge boson loops. However, in Ref. [1], it has been shown that the latter contribution is negligibly small compared to the former one. Therefore, we consider only the top-sector contribution in order to obtain the constraint. For the detailed expression of the top-sector contributions, see Ref. [19]. In Fig. 2, constraints on m T and m AH at 68% and 99% confidence levels are depicted. At each point in these figures, the Higgs mass is determined to satisfy the WMAP constraint on the U branch (left panel) and L branch (right panel). To obtain the constraint, we have used three experimental values: the W boson mass (m W 8:412 :42 GeV), the weak mixing angle (sin 2 lept eff :231 53 : 16) [21], and the leptonic width of the Z boson ( l 83:985 :86 MeV) [22]. We have also used the fine structure constant at the Z pole [ 1 m Z 128:95 :48] and the top quark mass (m t 172:7 2:9 GeV) [23]. As seen in these figures, the lower bound of the T quark mass is about 6 GeV, and the mass difference between m T and m AH is larger than 5 GeV. 1 In the excluded region (solid black), there is no combination of the parameters 1, 2, and f which can give the correct top quark mass via Eq. (14). The parameter region with large m AH corresponds to the region with heavy m h due to the WMAP constraint. The heavy Higgs contribution to the T parameter can be canceled by those from the top partner, T, if its mass is tuned appropriately. When m AH m h =2 increases, the cancellation can be achieved only for a small region of m T, as we can see in these figures. IV. T QUARK PRODUCTION AT THE LHC In this section, we study the signature of T quark production at the LHC. Following the discussion in the previous section, we first discuss properties of the T quark and present a few representative points used in our simulation study. We next consider a top reconstruction for the 1 As shown in the previous subsection, the dominant annihilation mode of dark matter for the relic abundance is A H A H! W W, implying that m AH >m W. signal using a hemisphere analysis. Finally, we address the background to this process, which comes from SM t t production. A. Properties of the T quark Because of the T-parity conservation, the T quark would be pair produced from proton-proton (pp) collisions at the LHC. Since the T quark has a color charge, it is produced dominantly through SU 3 c interactions at the LHC. The production cross section depends only on its mass, m T. Unlike the scalar top in the MSSM, the T quark is a Dirac fermion. Hence its production cross section at the LHC ranges from.1 1 pb when m T is less than 1 TeV as shown in Fig. 3 (left panel). The decay process of the T quark is simple, because only A H and Z H (W H ) are T-odd particles lighter than the T quark. Furthermore, the T quark is almost an SU 2 L singlet. The interactions relevant to the decay are L i 2g 5 cos HT A6 H sin P L sin P R t i 2g 5 sin HT Z6 H sin P L sin P R t: (16) As seen in Eq. (16), the decay mode T! Z H t is highly suppressed by sin 2 H ; therefore T decays dominantly into the stable A H and the top quark. In Fig. 3 (right panel), the decay width is depicted as a contour plot in the m T ;m AH plane. It shows that the width is typically several GeV. The signal of T quark production is a top pair (t t) with significant missing transverse momentum. In this paper, we assume that the branching ratio of the process T! A H t is 1%. For simplicity we also assume that the other extra matter fermions do not contribute to the signal. The model points we choose for the simulation study are listed in Table II. The production cross sections are also shown in the table, which were obtained by the COMPHEP code [24] using the CTEQ6L1 parton distribution function [25] and the scale of the QCD coupling set to be Q m T. In order to generate parton level events, we calculated the process pp! t ta H A H directly using the COMPHEP 556-5

MATSUMOTO, NOJIRI, AND NOMURA PHYSICAL REVIEW D 75, 556 (27) σ( pp T T X) (pb) 1 1.1.1 m (GeV) AH 3 2 1 2 GeV 1 GeV 3 GeV 4 GeV 5 GeV 5 1 15 m (GeV) T 6 1 14 m (GeV) T FIG. 3. Production cross section of the T quark at the LHC as a function of m T (left panel). Contour plot of the T decay width in the m T ;m AH plane (right panel). code, keeping diagrams relevant to the process through onshell T T production. We generated 1 events for each model point to study their distributions. The parton level events were interfaced to HERWIG [26] for fragmentation, initial and final state radiations, and hadronization. The effect of the top polarization is not included in our simulation. The detector effects were simulated by the ACERDET code [27]. This code provides a simple detector simulation and jet reconstruction using a simple cone algorithm. It also identifies isolated leptons and photons, finds b and jets, and calculates the missing momentum of the events using calorimeter information. Before going to the discussion of top reconstruction, we define two important quantities which are frequently used for new physics searches at the LHC. One is the missing transverse energy, E Tmiss P 2 Tx P2 Ty 1=2, which is important for the signal of the models with a stable dark matter candidate such as the MSSM or the little Higgs model with T parity. Here P Ti is the sum of transverse momenta measured by the calorimeter. The other is the effective transverse mass, M eff X X X T p T p T E Tmiss ; jetsp isolated leptons isolated photons (17) where we require that the pseudorapidity is less than 3 ( <3) and the transverse momentum is p T > 5 1 GeV for each jet (lepton/photon). The effective transverse mass is a good quantity to measure the mass scale of a produced particle. TABLE II. The model points for our simulation study. The production cross section for the T quark is also shown. m T (GeV) m AH (GeV) pp! T T X (pb) I 6 1.94 II 7 125.382 III 8 15.171 IV 9 175.822 B. Top reconstruction Now we move on to the discussion of top quark reconstruction for the signal. The top quarks produced from T quarks have significant p T. Since the T quarks must be produced in a pair, we expect two separate jet systems originating from the two top quarks. We therefore use the hemisphere analysis [28] for the event reconstruction. Two hemispheres are defined in each event, and high p T jets, leptons, and photons are assigned to one of the hemispheres. Specifically, (i) Each hemisphere is defined by an axis P i (i 1, 2), which is the sum of the momenta of high p T objects ( jets/leptons/photons) belonging to hemisphere i. Only the jets with p T > 5 GeV and leptons/photons with p T > 1 GeV are assigned to the hemisphere in order to reduce the contamination of QCD activity. (ii) High p T objects k belonging to hemisphere i satisfy the following condition: d p k ;P i <d p k ;P j ; (18) where i and j are the indices of the hemispheres, and the function d is defined as d p k ;P i E i jp i j cos ik E i E k 2 ; (19) where ik is the angle between P i and p k. To find the axis satisfying the above conditions, we take the following steps. (1) We take the highest p T object i (jets/leptons/photons) with momentum p i, and the object j with the largest Rjp j j, where R i; j 2 i; j 2 1=2. We take p i and p j to be the seeds of the hemisphere axes, namely, P in 1 p i, P in 2 p j. (2) Each object with momentum p k is assigned to the hemisphere i, ifd p k ;P in i <d p k;p in j. (3) We then define new Pin i (i 1, 2) as the sum of the momenta of the objects belonging to the hemisphere i. (4) We repeat processes (2) and (3) until assignment converges. In this paper, we E i 556-6

HUNTING FOR THE TOP PARTNER IN THE LITTLEST... PHYSICAL REVIEW D 75, 556 (27) 3 16 2 1 12 8 4 1 2 3 4 5 m(jj) (GeV) 1 2 3 4 5 m(jjj) (GeV) FIG. 4. Distribution of a two jet invariant mass, m max jj (left panel), and that of m min jjj (right panel) in the pp! T T! t ta H A H process for point III. denote the hemisphere seeded from the highest p T object as hemisphere 1 and the other as hemisphere 2. If there are more than two jets in a hemisphere, we can calculate the maximum of the two jet invariant masses for all combinations of jets in the hemisphere, m max jj. In Fig. 4 (left panel), we plot this for hemisphere 1 using point III in Table II. The distribution has two peaks around m jj 8 GeV and 13 GeV. The lower peak corresponds to the jet pair arising from the W decay, while the second peak comes from the combination of one of the two jets from a b quark and one of the partons from the W decay. When a hemisphere contains more than three jets, we can also define the minimum three jet invariant mass in the hemisphere, m min jjj, where two of the three jets are those which give m max jj. In Fig. 4 (right panel), the distribution peaks at the input top quark mass 175 GeV, clearly showing that the hemisphere analysis can group the jets from the top quark correctly. In the following, we often require the top mass cut, m min jjj < 2 GeV, for at least one of the hemispheres. The efficiency to find at least one top quark candidate in a signal event is moderate, about 2%. In Fig. 4,wehavea long tail at m min jjj > 2 GeV, which consists of events with additional jets or misassignment of jets. If we optimize the top search strategy after the hemisphere reconstruction, we may increase the top reconstruction efficiency, but we do not study this possibility in this paper. It should be noted that the reconstructed m min jjj distribution in the hemisphere analysis is not biased, because we do not assume the existence of the top quark in the event. C. Background The most important background comes from the pp! t t X process in the SM. The tree-level production cross section of the top quark is 4 pb. 2 We have generated 2 The NLO cross section is 8 pb [29]. events corresponding to 5 fb 1 for this study. The cross section after very weak cuts, M eff > 4 GeV, E Tmiss > 1 GeV, n jets p T > 1 GeV n 1 1, and n jets p T > 5 GeV n 5 2, is about 16 pb, which is still higher than the signal cross section by more than a factor of 1. The efficiency to find the top quark candidate in a hemisphere becomes lower for the t t production process if large E Tmiss is required. This is because the W boson from the top quark decay must decay leptonically to give such a high E Tmiss to the event. This can be seen in Table III, where we have listed the probability to find three jets with m min jjj < 2 GeV in one, or both, of the hemispheres. In Fig. 5, we present the m min jjj distribution in a two dimensional plot, where the x (y) axis corresponds to m min jjj in hemisphere 1 (hemisphere 2). While the distribution peaks around m t in both of the hemispheres for the signal (left panel), the distribution for t t production scatters over the plot (right panel). This means that at least one of the top quarks has to decay leptonically, and therefore some of the jets must be additional QCD jets in t t production. If we require a top candidate in both hemispheres, the t t background reduces to the 3 fb level; however, the reconstruction efficiency for the signal also decreases due to additional jets in the final state and the overlap of jets. In the next section, we look for the excess of signal events over the t t background for the events with m min jjj < 2 GeV in at least one of the hemispheres, for the various sample points. In this case, the t t background cross section is around 16 pb :153 2:5 pb, which is many orders of magnitude larger than any other irreducible backgrounds listed in [13]. In Ref. [3], it has been shown that the quark production is not the dominant background in the inclusive study of SUSY processes for the -lepton E Tmiss channel. The backgrounds from W and Z boson production are about as important as that due to t t production. However, one should be able to reduce these backgrounds significantly by 556-7

MATSUMOTO, NOJIRI, AND NOMURA PHYSICAL REVIEW D 75, 556 (27) TABLE III. Fraction of T T and t t events with m min jjj < 2 GeV in one of the hemispheres, and in both of the hemispheres. The probability to reconstruct the top quark in both of the hemispheres is significantly small for the t t production. pp! T T pp! t t m min jjj 1 < 2 GeV or m min jjj 2 < 2 GeV 22.9% 15.3% m min jjj 1 < 2 GeV and m min jjj 2 < 2 GeV 1.6%.17% 5 5 m2(jjj) (GeV) 4 3 2 1 Signal 3 2 1 m2(jjj) (GeV) 4 3 2 1 Background 8 6 4 2 1 2 3 4 5 m1(jjj) (GeV) 1 2 3 4 5 m1(jjj) (GeV) FIG. 5 (color online). Reconstructed m min jjj distributions for pp! T T for point III (left panel) and pp! t t (right panel) processes. We take E Tmiss > 4 GeV and n 1 1 and n 5 2 for the t t events. The efficiency to find the top candidate in each hemisphere [m min jjj i < 2 GeV] is low in the t t production compared to the signal process. requiring a reconstructed top and b tagged jets. We therefore do not study these in this paper. V. DISCOVERY OF THE T QUARK AT THE LHC In this section, we investigate the possibility of finding the T quark at the LHC. First, we discuss the separation of the signal from the background using their different kinematic properties. We find the region in the E Tmiss and M eff plane where the signal best dominates the background. Next, we calculate the statistical significance for T quark discovery at the LHC, and show that the significance exceeds 7 for all sample points. We also calculate the M T2 variable [31] to investigate the possibility of extracting information about m T from the signal. Finally, we comment on the differences in production and decay distributions between T and scalar top signals. A. Separation of signals from backgrounds In order to separate T T signals from t t backgrounds, we further study E Tmiss distributions for a given (high) M eff interval. In Fig. 6, we show the signal and background distributions in the M eff and E Tmiss plane. The E Tmiss distribution of the signal peaks near its maximum ( :5M eff ) for significantly large M eff. This feature is common in processes where new particles are pair produced and each decays into a stable neutral particle and other visible particles [32]. This can be understood as follows. Since new heavy particles are produced mostly 8 Signal 1 8 8 Background 3 E Tmiss (GeV) 6 4 6 4 2 E Tmiss (GeV) 6 4 2 1 2 6 1 14 18 M eff (GeV) 2 6 1 14 18 M eff (GeV) FIG. 6 (color online). E Tmiss versus M eff distributions for the T T production at the LHC for point III (left panel), and the t t background (right panel). Normalizations in both figures are arbitrary. 556-8

HUNTING FOR THE TOP PARTNER IN THE LITTLEST... PHYSICAL REVIEW D 75, 556 (27) jets jets ν jets TT AH AH ν jets (T T production) (t t production) FIG. 7. T T production which gives the largest E Tmiss (left panel), and background t t production with neutrino emissions. near the threshold at the LHC, the velocity of the T quark is low in the transverse direction. Then, E Tmiss and M eff are maximized when the decay configuration of the two T quarks is such that the two top quarks from the decay go in the same direction in the rest frame of the T T system, as illustrated in Fig. 7. In this case, M eff 2E Tmiss 2m T. On the other hand, the kinematics is totally different for the background t t distribution. As shown in Fig. 7, neutrinos arising from top decays are collinear to the direction of the parent top quark, if the center of mass energy in the collision is much higher than the top quark mass. Therefore, E Tmiss :5M eff is kinematically disfavored. t t This can be seen in Fig. 6 (right panel). As M eff increases, the fraction E Tmiss =M eff is significantly reduced. Using the nature of these production and decay kinematics, we can find the kinematical region with good separation between signal and background. We restrict M eff to a certain large value, so that we see the bump of the signal event in the E Tmiss distribution. In Fig. 8, we show the signal and background E Tmiss distributions for points I to IV (top four panels). Each plot corresponds to the integrated luminosity R dtl 5 fb 1. We restrict M eff to 2m T 2 GeV <M eff < 2m T for m T 6 and 7 GeV, and 2m T 3 GeV <M eff < 2m T for m T 8 and 9 GeV, so that E Tmiss becomes maximal and the signal rate is still reasonably high. Here the top mass cut, m min jjj < 2 GeV, is required for at least one of the hemispheres, which reduces the background by a factor of 5 and the signal by a factor of 3 compared to the case where no cut is applied on m min jjj. In the bottom four panels in Fig. 8, we further require that the events have no isolated leptons. The isolated lepton is produced by the leptonic decay of the top quark. The t t background is reduced by a factor of 2 by this cut, with no significant reduction of signal events. Each distribution shows the clear excess of events over the (exponentially decreasing) background. The excess is less prominent for smaller M eff bins. We show the E Tmiss distribution for point II with 5 4 3 2 1 25 2 15 1 5 12 3 16 m T = 6 GeV m T = 7 GeV m T = 8 GeV m T = 9 GeV 12 8 4 3 5 7 3 5 7 3 5 7 9 3 5 7 9 E Tmiss (GeV) E Tmiss (GeV) E Tmiss (GeV) E Tmiss (GeV) 4 2 3 5 7 3 5 7 3 5 7 9 3 5 7 9 E Tmiss (GeV) E Tmiss (GeV) E Tmiss (GeV) E Tmiss (GeV) 8 4 No isolated lepton 8 No isolated lepton No isolated lepton No isolated lepton 5 2 m T = 6 GeV m T = 7 GeV m T = 8 GeV m T = 9 GeV 6 4 15 3 2 1 2 1 1 5 FIG. 8 (color online). E Tmiss distributions for points I to IV from left to right, with 2m T 2 3 GeV <M eff < 2m T for m T 7 8 GeV. The red (dark gray in the black-and-white version) and blue (darkest gray) histograms are for the signal and background distributions, and the green (light gray) histograms are the sum of these. m min jjj < 2 GeV is required in at least one of the hemispheres. In the bottom four panels, only the events without isolated leptons have been used. 556-9

MATSUMOTO, NOJIRI, AND NOMURA PHYSICAL REVIEW D 75, 556 (27) 5 4 3 2 1 25 2 15 1 5 1 TeV < M < 1.2 TeV eff 5 m T = 7 GeV 4 m T = 7 GeV 3 2 1 3 5 7 3 5 7 9 E Tmiss (GeV) E Tmiss (GeV) No isolated lepton No isolated lepton 1 TeV < M eff < 1.2 TeV 3 1.4 TeV < M eff < 1.6 TeV m T = 7 GeV m = 7 GeV 2 1 1.4 TeV < M eff < 1.6 TeV 3 5 7 3 5 7 9 E Tmiss (GeV) E Tmiss (GeV) T FIG. 9 (color online). E Tmiss distributions for point II with 1 GeV <M eff < 12 GeV (left two panels) and 14 GeV <M eff < 16 GeV (right two panels). m min jjj < 2 GeV is required for at least one of the hemispheres. 1 GeV <M eff < 12 GeV in Fig. 9 (left two panels). While the number of the signal increases by a factor of 2, the t t background increases by more than a factor of 3. On the other hand, if one increases M eff, the number of the signal reduces rather quickly, and we find that the events in this region do not contribute much to the discovery of the T quark. See Fig. 9 (right panel) for the distribution with 14 GeV <M eff < 16 GeV. Note that the events in the region M eff 2m T arise from highly boosted T quarks; therefore the fraction E Tmiss =M eff decreases, making the signal distribution less prominent over the background. The numbers of the signal and background events in the signal region are shown in Table IV. Here we take the same signal region as that of Fig. 8. The signal to background ratio with the top cut is more than 3 for all sample points. Thus, it is clearly shown that the signal dominates in the region where E Tmiss :5M eff. B. Statistical significance for the T quark discovery As shown in Fig. 8, the t t background reduces rather quickly at large E Tmiss. Near the end of the distribution of E Tmiss, the distribution is dominated by the signal in Fig. 8. On the other hand, the signal distribution decreases quickly after its peak. We therefore fit the decrease of total distribution near the maximum value :5M eff to the Gaussian TABLE IV. The signal to background ratio at sample points I to IV. We take a region with Meff min The ratio is best if one requires the top cut and vetoes isolated leptons. <M eff < 2m T and E Tmiss >E cut Tmiss. m T (GeV) Meff min (GeV) E cut Tmiss (GeV) Signal/BG (-lepton with top cut) Signal/BG (with top cut) Signal/BG 6 1 4 842/16 153/313 3336/134 7 12 45 263/54 332/114 1284/582 8 13 5 28/28 249/57 874/417 9 15 55 93/7 15/16 397/23 556-1

HUNTING FOR THE TOP PARTNER IN THE LITTLEST... PHYSICAL REVIEW D 75, 556 (27) TABLE V. Fit of the E Tmiss distribution to the Gaussian function near :5M eff. Here, N bin is the number of bins used for the fit and E bin is the bin size. m T (GeV) h h (GeV) N bin E bin (GeV) h= h 6 19 14.7 38.5 7 2.8 13 7 16 6.8 43 6 25.6 24 8 46 4.9 66 6 3.4 9.4 9 17 2.4 76 6 35.2 7.1 function, F E Tmiss hexp E Tmiss E avg 2 2 Tmiss 2 : (2) The result is summarized in Table V, where we have used six or seven bins with the bin size ( E bin ) between 2.8 and 35.2 GeV. These fits give 2 = N bin 3 1. The statistical significance of the signal is given by h= h, where h is the height of the Gaussian distribution and h is its error. The significance is more than 7 for all sample points. When we fit the background distribution to the same Gaussian function, we obtain t t as 12 (125) GeV for the fit above E Tmiss > 34 328 GeV for 1 GeV <M eff < 12 GeV (12 GeV <M eff < 14 GeV), respectively. The 2 of the fit is 2 = N bin 3 1:1 (1.2). We find t t ; therefore, the detected edge is clearly inconsistent with the t t background distribution. We found that the signal is prominent over the background when M eff is restricted to the region near 2m T.We therefore calculate the M T2 variable for the events near the bump. We use Cambridge M T2, which is defined as MT2 2 min max m 2 p A H 1T p A T H p vis 1 ;pa H 1 ;m2 T p vis 2 ;pa H 2 : 2T P Tmiss (21) It is a function of the transverse momenta and masses of the two visible particles, P Tmiss, and the mass of the invisible particle. M T2 is sensitive to the mass difference m T m AH, but is not sensitive to the overall mass scale. We take events where there are more than three jets with p T > 5 GeV in each hemisphere. We calculate M T2 by taking the visible particle momenta as the sum of the three jet momenta used to calculate the minimum three jet mass m min jjj of the hemispheres, and we fix m AH 15 GeV. We also require m min jjj < 2 GeV for at least one of the hemispheres. The distributions are shown in Fig. 1. We found a positive correlation to the mass of the top partner; however the number of events that can be used for this analysis is small. C. Difference between T and scalar top signals We discuss the differences in the production and decay distributions between the T quark and scalar top (~t) signals. It is impossible to find the process pp! ~t~t followed by the decay ~t! t~ 1 at the LHC, because the production cross section is a factor 1 smaller than the T quark production cross section. However, we can still learn something from the comparison. Here we take m T m ~t 8 GeV and m AH m ~ 1 15 GeV. We took low energy parameters of the MSSM so that all particles except stop and the LSP ( ~ 1 ) are too heavy to be accessible at the LHC, and ~t decays into t and ~ 1 with a 1% branching ratio. The events for the pp! ~t~t! t t~ 1 ~ 1 process are generated by the COMPHEP code and interfaced to HERWIG. 4 m T = 6 GeV 1 TeV < M eff < 3 TeV E >.4TeV Tmiss 16 m T = 7 GeV 1.2 TeV < M eff < 3 TeV E >.45TeV Tmiss 8 m T = 8 GeV 1.3 TeV < M eff < 3 TeV E >.5TeV Tmiss 3 12 6 2 8 4 1 4 2 3 5 7 9 3 5 7 9 3 5 7 9 M T2 (GeV) M T2 (GeV) M T2 (GeV) FIG. 1 (color online). Distribution of the M T2 variable for the events in the signal region defined in Table IV. m min jjj < 2 GeV is required in at least one of the hemispheres. m T 6, 7, 8 GeV from left to right. 556-11

MATSUMOTO, NOJIRI, AND NOMURA PHYSICAL REVIEW D 75, 556 (27) In Fig. 11 (left panel), we show the p T distributions of T and ~t production at the LHC. The cross section is dominated by the low p T component for the T quark. On the other hand, the ~t production cross section is dominated by the P wave; therefore the p T distribution is broadened. The production cross sections peak at 35 and 6 GeV for the T quark and ~t, respectively, where :4 and.6. Both cross sections are kinematically suppressed near the peak by. Especially in the stop case, the squared amplitude is proportional to 2 due to the P wave production. The cross section beyond the peak is suppressed by the quickly decreasing quark and gluon parton distribution functions. If this p T distribution is reconstructed from the t t distribution, this indirectly suggests that the produced particle is a fermion. The difference of the distributions affects the E Tmiss distribution for fixed M eff as shown in Fig. 11 (center and right panels). Here, the center panel shows the event distribution with 13 GeV <M eff < 16 GeV, and, in the right panel, when the top cut is required for at least one of the hemispheres. The E Tmiss distribution of the T events has a sharper peak near :5M eff, while the distribution is broader for ~t. This is because ~t is produced with higher p T ; therefore the top quark has, on average, a larger energy. Therefore the average fraction E Tmiss =M eff becomes smaller. Finally we comment on the top reconstruction for the supersymmetric signature. In supersymmetric models, the top quarks mostly arise from gluino decays through processes ~g! ~tt, ~g! ~bb! tb~. The top quark in the event is hardly seen in the m min jjj distribution due to the additional high p T jets and leptons. In previous studies, the top quark is therefore searched for by looking for the jets consistent with the top decay kinematics, namely, m jj 8 GeV and m bjj 175 GeV by looking for the combination of jets i, j, and k, which minimizes 2 defined as 2 m i; j m W 2 m W 2 m i; j; k m t 2 m t 2 ; (22) where m W and m t are the expected errors on W and top mass reconstruction. The efficiency to find the top quark in the events becomes higher if we look for the jet combination consistent with top decay kinematics; however, in this case, we occasionally find top candidates in events which do not contain parton level top quarks, increasing backgrounds from the other SM processes [33]. The signal distributions may also be distorted by such a procedure. This is particularly the case when the number of jets in the events is large. A scheme to subtract the accidental background from the top decay kinematical distributions has been developed [12]. VI. DISCUSSION AND OUTLOOK In this paper, we have studied the phenomenology of the top partner in the littlest Higgs model with T parity. This model predicts a stable neutral massive gauge boson (A H ), and a relatively light top partner (T ). The EW precision measurements and dark matter observations constrain parameters of the model, f, m h, 1, and 2. We find m AH and m h are strongly constrained by the relic density constraint, if coannihilation processes with other T-odd particles are not efficient. Combined with the EW precision measurements, we have found the lower limit of m T as a function of f, and clarified the fact that the top partner mass must be above 6 GeV, and the mass difference between T and A H is large. The signature of T quark pair production at the LHC is therefore high p T top quarks from T! ta H 3 2 6 4 Without top cut 2 With top cut 1 2 1 4 8 12 16 2 p (GeV) T 3 5 7 9 E Tmiss (GeV) 3 5 7 9 E Tmiss (GeV) FIG. 11. The p T distribution of T (solid line) and ~t (dashed line) at the LHC (left panel). The 1 5 events are generated by the COMPHEP code. We set m T m ~t 8 GeV. The E Tmiss distribution for T (solid line) and ~t (dashed line) productions with 13 GeV <M eff < 16 GeV (center and right panels). For the right panel, we required m min jjj < 2 GeV for at least one of the hemispheres. 556-12

HUNTING FOR THE TOP PARTNER IN THE LITTLEST... PHYSICAL REVIEW D 75, 556 (27) and missing p T coming from the A H. We have studied the signal and background distributions at the LHC. Unbiased reconstruction of the top quark is essential to establish the existence of the top partner without increasing accidental backgrounds. We apply the hemisphere analysis, which is an algorithm to find two axes in events originating from particles produced in pairs. We have found that the appropriately defined three jet invariant mass in a hemisphere shows a clear peak at the top quark mass with a tail due to misreconstructed events. The three jet invariant mass is calculated from the pair of jets giving the maximum invariant mass in a hemisphere and a jet giving the minimum three jet invariant mass when combined with this jet pair. The efficiency to reconstruct at least one top quark, i.e. a hemisphere with m min jjj < 2 GeV in T T events, is reasonable, 2%. The dominant background to the T T process is t t production, whose production cross section with E Tmiss > 4 GeV is O 1 or more times larger than the signal cross section. We study the E Tmiss distribution as a function of M eff. Because of a simple kinematic reason, the separation of the signal from background is best when M eff 2m T. A pseudoedge structure of the signal can be observed over the t t background in that region, and the significance of the signature turns out to be larger than 7 for the case of m T 9 GeV. The reconstructed signal is clearly different from that of SUSY events. In the case of SUSY, the production cross section of ~t~t is rather small and is not detectable. Scalar top quarks may be produced from gluino decay. In that case, additional jets are also produced, and the top cut m min jjj < 2 GeV for the jets in a hemisphere can hardly be satisfied. Results of our simulation given in this paper should be regarded as an order of magnitude estimate. We have used the simple smearing and jet finding algorithms provided by the ACERDET code. The E Tmiss and jet energy resolution could be different at the ATLAS and the CMS detectors in the LHC environment. Note that the energy resolution assumed by ACERDET is closer to p the ATLAS detector [The default is E=E 5%= p E, while E=E 5 1 %= E for the ATLAS (CMS) detector]. Furthermore, the background from t t production is estimated by the HERWIG code in this paper, where t t 4 pb. However, it is well known that the t t cross section receives large NLO corrections of around a factor of 2. The multijet final state t t jets should also give an important background to the events with E Tmiss. For the parameters we have studied, S=N is high in the signal region. Moreover, the signal has a peak structure in the E Tmiss distribution. Therefore, we believe that our result is stable against additional sources of background and a factor of 2 increase of the background. In addition, t t production with a multijet final state has a better chance of having a large three jet invariant mass in a hemisphere, because events have additional jets which are not kinematically constrained by the top mass shell conditions. It is likely that the process t t jets would be rejected by the top cut m min jjj < 2 GeV. It would be interesting to study the processes t t jets and T T jets to check this expectation explicitly. In this paper, we do not study mass reconstruction in detail. The mass difference m T m AH could be extracted from the M eff E Tmiss distribution if the background distribution can be calibrated precisely. We have also looked into the M T2 distribution for the sample with at least one reconstructed top quark, which shows positive correlation with the mass difference. The fit may be improved if the efficiency to reconstruct the top quark is improved by selecting the jets consistent with the top decay kinematics. The combinatorial backgrounds may be reduced if we select the top candidate among the jets in a hemisphere. The existence of the top partner and dark matter associated with the T-parity symmetry is an important feature of physics beyond the standard model, as we discussed in the Introduction. The collider phenomenology of the top partner therefore deserves more realistic studies. We hope more realistic studies will be performed by the ATLAS and CMS groups in the future. ACKNOWLEDGMENTS This work is supported in part by the Grant-in-Aid for Science Research, Ministry of Education, Culture, Sports, Science and Technology, Japan (No. 1681211 for S. M., and No. 168127 and No. 18346 for M. M. N.). [1] ATLAS Collaboration, ATLAS Detector and Physics Performance Technical Design Report, Report No. CERN-LHCC-99-14/15. [2] CMS Collaboration, Physics TDR, Vol. I, Report No. CERN-LHCC-26-1, 26; Physics TDR, Vol. II, Report No. CERN-LHCC-26-21, 26. [3] S. P. Martin, hep-ph/979356. [4] I. Hinchliffe, F. E. Paige, M. D. Shapiro, J. Soderqvist, and W. Yao, Phys. Rev. D 55, 552 (1997); H. Bachacou, I. Hinchliffe, and F. E. Paige, Phys. Rev. D 62, 159 (2). [5] N. Arkani-Hamed, A. G. Cohen, and H. Georgi, Phys. Lett. B 513, 232 (21); N. Arkani-Hamed, A. G. Cohen, E. Katz, A. E. Nelson, T. Gregoire, and J. G. 556-13